Real barriers, not just rule hits
A checker finds a missing attribute. Tallyn reads the page and finds the barrier: the clickable div that traps a keyboard, the alt that should be empty, the control with no accessible name.
AI accessibility auditing for web teams
Find what makes
your site hard to
use, and fix it.
Paste a URL or your HTML. Tallyn finds the real accessibility barriers, ranks them by impact, explains who each one affects, and drafts the exact code fix.
Free for your first page every month. A strong aid, not a legal certification.
signup.html
scanning
<form class="signup" action="/join">
<h2>Create your account</h2>
<input type="email"
placeholder="Email address">
<img src="/badge.png">
<div class="btn" onclick="join()">
Sign up
</div>
<a href="/plans">click here</a>
</form>
The problem
Most sites fail real users
Accessibility is not an edge case. A large share of people hit a barrier on an ordinary page, most barriers are invisible to the team that shipped them, and the people they block usually leave without a word.
- 96%
- of the web's most-visited home pages have detectable WCAG failures
- 1 in 4
- adults in the US live with a disability
- ~60
- accessibility errors on an average home page, before anyone looks deeper
- 71%
- of users with a disability leave a site that is hard to use, rather than report it
Figures are illustrative, drawn from widely cited accessibility research such as the WebAIM Million and CDC disability data. Tallyn does not claim these exact numbers for your site. It measures yours.
How it works
Three steps, about forty seconds
01
Paste a URL or your HTML
Give Tallyn a public URL and it fetches the page, or paste the markup for something behind a login or built in the browser. Nothing to install, no config, no template to pick. Tallyn works out what kind of page it is reading.
A URL that returns real server-rendered markup works best. If a page ships an empty shell, Tallyn tells you it saw little to read rather than guessing.
02
Tallyn finds and ranks the barriers
Deterministic checks catch the clear failures, then a language model reasons about the page in context: a clickable div that traps a keyboard, an alt that should be empty, a control with no accessible name. Every finding is ranked by how much it actually blocks people.
Each quoted snippet is checked back against your own markup before you see it. A fix that points at the wrong element is worse than no fix, so an unlocatable one is dropped.
03
Get plain explanations and drafted fixes
The issues come back grouped by severity, each with who it affects, why it matters, the WCAG reference, and the exact code fix written for your own markup. Copy the fix, ship it, and re-check the page to watch the score move.
The fix is a strong first draft, not a patch to merge blind. Each one carries a one-line note on why it works, so you can adapt it with confidence.
See an audit
A real one, on a real page
This is a typical e-commerce checkout. Click any issue. Nothing here is a mockup: it is the product screen, with the product’s own output and the same copy-ready fix.
Checkout page
shop.example.com/checkout
54
score · At risk
2
critical
2
serious
5
issues total
Issues, ranked by impact
CriticalWCAG 2.1 · 1.3.1 Info and Relationships (A)Mechanically verified
The email field has no label
Who it affects
Screen reader users, who reach this field and hear only "edit text" with no idea it wants their email, and voice-control users, who have no name to speak.
Why it matters
The input relies on a placeholder for its name, and a placeholder is not a label: it disappears on focus and is not reliably announced. At checkout, a field a person cannot identify is a field they cannot complete, so the sale is lost at the last step.
In your markup
<input type="email" placeholder="Email address">The fix
<label for="email">Email address</label>
<input id="email" type="email" name="email"
autocomplete="email">A real <label> tied to the input by id gives it a name for every assistive technology, and autocomplete speeds up everyone else.
The read
This is a standard e-commerce checkout, and two of its issues block people outright: the email field has no label and the pay control cannot be reached by keyboard. Fix those two first, then the low-contrast total and the missing product alt text. The structure underneath is sound.
What you get
Six things, all of them the point
Ranked by impact
Forty findings help nobody. Tallyn orders them by how much they block real people, so you fix what matters first.
- Critical
- blocks the task
- Minor
- worth cleaning up
Plain-language impact
Every issue names who it affects and what it does to them, so it is a real consequence, never a bare rule id.
“Screen reader users hear only ‘edit text’ and cannot tell what to type.”
Drafted code fixes
Each issue comes with the corrected markup, written for your own content, ready to copy.
<label for="email">Email</label> <input id="email" type="email">
Re-check after you ship
Run the page again and see which issues cleared and how the score moved. Progress you can show the team.
Team and Business plans
Model-agnostic routing
Each step goes to the model that is best at it, and a large or difficult page escalates automatically.
See the routing tableBuilt AI-native
The model is the engine, not a feature we added
Tallyn was not a rules engine that later added AI. There is no version of this product without modern language models, because the work is reasoning. No rule can tell whether an alt should be empty or should describe a chart, whether a clickable div traps a keyboard, or which three of forty findings actually block someone.
Tallyn is model-agnostic by design. Every step asks for a task, not a model, and one file decides which lab answers. A lab changing its price, its policy or its quality is a configuration change here, not a rebuild. Large or structurally difficult pages escalate to a frontier model on their own.
Routing table, generated from the code
- Read the page and find the real accessibility barriersgpt-4.1-mini/balanced
- Draft the exact code fixgpt-4.1-mini/balanced
- Write the overall read for the reportgpt-4.1-mini/balanced
- Work out what kind of page this isgpt-4.1-nano/fast
Candidates from other labs are declared behind the same interface and switch on when their key is present. The escalation rule and the fallback chain are on the How the AI works page.
Where we draw the line
A strong aid, not a legal certification
Tallyn finds the real barriers on a page, explains who they affect, and drafts the fix. That gets you most of the way and makes the rest cheaper. It is not a legal certification of compliance, and no automated tool can be one, because some accessibility questions need a person testing with a screen reader and a keyboard.
Use Tallyn to fix the clear problems fast, then bring a manual audit in for sign-off. We say this plainly, and the same line is on every report.
Read our stance on care and privacyWeb teams
What it catches
“We had passed our automated checks and still failed a manual audit on the two things that mattered most: an unlabeled search field and a menu you could not open with a keyboard. Tallyn found both in one pass, told me who they blocked, and handed me the fix. We shipped it that afternoon.”
“Every new page goes through Tallyn before it ships now. I get the three issues to fix instead of a forty-item report, ranked by who they actually block, with the code already written. Our accessibility backlog stopped growing for the first time in two years.”
Illustrative stories, written to show what Tallyn catches. They are composites, not quotes from named customers.
Pricing
One page free, every month
Prices are in US dollars and that is the only price list there is. No setup fee, no minimum term, cancel from the dashboard.
MonthlyAnnual
Switch to annual and two months are free.
Free
One page a month, audited properly. Enough to find out where a page fails real users and get the fixes.
$0forever
No card required
1 page audit a month
- 1 page audit a month
- Every real barrier found and ranked by impact
- Plain-language explanation of who each issue affects
- The exact code fix, ready to copy
- Paste a URL or your HTML
Team
For the team shipping every week. Audit as many pages as you need, re-check after a fix, and export the report.
$29per month, USD
$348 a year, or $288 on annual
Start on Team
Card payments are not switched on in this deployment yet, so this joins the list and we email you the moment it is.
Up to 500 page audits a month, re-checks, export
- Everything in Free
- Up to 500 page audits a month
- Re-check a page after you ship the fix
- Audit history you can search
- Export a report as HTML or PDF for the team
- Every drafted code fix, on every issue
- Email support
Business
For the org that owns many sites. Whole-site scans, seats for the team, and answers the same day.
$99per month, USD
$1,188 a year, or $984 on annual
Start on Business
Card payments are not switched on in this deployment yet, so this joins the list and we email you the moment it is.
Whole-site scans, 5 seats, priority support
- Everything in Team
- Whole-site scans across many pages at once
- 5 seats, with a shared audit library
- Standards your org holds every page to, applied to every audit
- A trend line as scores improve over time
- Priority support with a same-day first reply
Audit your next page before you ship it
One page a month, free, with every barrier ranked, explained, and the fix written out. No card.